The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes well-known, official Python packages from the Python Package Index (PyPI), including google-ads, gspread, and google-auth. These are standard dependencies for interacting with Google's Advertising and Workspace APIs. Additionally, it suggests cloning the vendor's repository from GitHub to access supporting scripts.
[COMMAND_EXECUTION]: The agent uses the Bash tool to run a local diagnostic script (scripts/run_diagnostic.py). This script processes account data and generates a structured report. The command-line interface uses standard argument parsing to handle user input like Customer IDs and pacing thresholds.
[DATA_EXFILTRATION]: The skill is designed to read Google Ads data and can optionally export findings to a user-specified Google Sheet. This data flow is central to the skill's purpose and uses official, authenticated Google API channels.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface as it processes external data from the Google Ads API (e.g., ad headlines and campaign names).
Ingestion points: Data enters the agent's context via GAQL query results fetched in scripts/run_diagnostic.py.
Boundary markers: The findings are presented in a structured console report and CSV file.
Capability inventory: The agent has access to Bash and Read tools.
Sanitization: The script extracts specific fields and applies internal logic to generate verdicts. Furthermore, the rules.md documentation provides clear instructions to the agent to interpret results only and defer all mutations to a human operator.

account-diagnostic